Optimized scheduling of cascade hydropower stations with advance risk control in dynamic operations

Abstract

The uncertainty in forecasting runoff can lead to operational scheduling risks in the scheduling of cascade hydropower stations, potentially impacting power generation efficiency and supply quality. This study proposes an optimized method for formulating scheduling decisions by implementing advance risk control in the dynamic scheduling process for cascade hydropower stations. Firstly, the improved VMD method is proposed to reduce the noise in forecasting runoff errors, followed by an LSTM model to predict these errors, enabling the correction of the forecasted runoff. Next, the CVaR method is utilized to dynamically quantify the risks of insufficient power generation and water surplus associated with the scheduling strategy of cascade hydropower stations. Finally, an optimized scheduling model is established to formulate a scheduling strategy that considers both power generation benefits and scheduling risks. A case study in the Wujiang River Basin demonstrates that the forecasting runoff correction method effectively reduces the Mean Relative Error (MRE) in forecasting runoff. The proposed optimized scheduling model improves the accuracy of scheduling decisions for the Dahuashui and Geliqiao hydropower stations during flood seasons by 1.04% and 0.17%, respectively, and reduces actual water surplus by 1.18% and 0.28%. In dry seasons, it increases the accuracy of scheduling decisions by 6.21% and 8.48%, respectively. This model ensures power generation efficiency while reducing operational scheduling risks and enhancing decision accuracy across varying seasonal conditions.